Visual Relationship Detection with Internal and External Linguistic Knowledge Distillation

ICCV 2017  ·  Ruichi Yu, Ang Li, Vlad I. Morariu, Larry S. Davis ·

Understanding visual relationships involves identifying the subject, the object, and a predicate relating them. We leverage the strong correlations between the predicate and the (subj,obj) pair (both semantically and spatially) to predict the predicates conditioned on the subjects and the objects. Modeling the three entities jointly more accurately reflects their relationships, but complicates learning since the semantic space of visual relationships is huge and the training data is limited, especially for the long-tail relationships that have few instances. To overcome this, we use knowledge of linguistic statistics to regularize visual model learning. We obtain linguistic knowledge by mining from both training annotations (internal knowledge) and publicly available text, e.g., Wikipedia (external knowledge), computing the conditional probability distribution of a predicate given a (subj,obj) pair. Then, we distill the knowledge into a deep model to achieve better generalization. Our experimental results on the Visual Relationship Detection (VRD) and Visual Genome datasets suggest that with this linguistic knowledge distillation, our model outperforms the state-of-the-art methods significantly, especially when predicting unseen relationships (e.g., recall improved from 8.45% to 19.17% on VRD zero-shot testing set).

PDF Abstract ICCV 2017 PDF ICCV 2017 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Visual Relationship Detection VRD Phrase Detection Yu et. al [[Yu et al.2017a]] R@100 29.43 # 1
R@50 26.32 # 1
Visual Relationship Detection VRD Predicate Detection Yu et. al [[Yu et al.2017a]] R@100 94.65 # 1
R@50 85.64 # 3
Visual Relationship Detection VRD Relationship Detection Yu et. al [[Yu et al.2017a]] R@100 31.89 # 1
R@50 22.68 # 1

Methods


No methods listed for this paper. Add relevant methods here